GPU Temperature Spikes + Crashing

ghost616

Reputable
Nov 5, 2014
16
0
4,510
At this point, I'm thoroughly convinced its a GPU issue, but I wanted other people's thoughts before I go get a new one:

Specs:

Phenom II X4 955 3.2 ghz
MSI 880GMS-E41(FX)
Kingston Hyper-X 1x8 GB DDR3
Palit GeForce GTX 660 OC
Intel SSD 335 180 GB + 2 Seagate HDDs
SilverStone 750W modular PSU (ST75F-GS)

Symptoms are as follows: at first, the computer monitor shuts off, cycles through analog/digital, but the computer is still on. Restarting the computer helps. Sometimes the whole PC freezes up. This only happens when playing 3D intensive games (BF4, DOTA2 so far). Didn't test other games as much as I didn't see the point. Now the crashing/monitor shutoff was completely random and therefore frustrating: sometimes, I can play for hours straight without crashing, but the next day, it starts crashing/monitor shutting off once or twice again in a day. NO error logs in Windows regarding the graphics, except for the improper shutdown logs. Again, the temperatures as per GPU-Z/MSI Afterburner top out at around 78C max, averaging 71C all throughout playing BF4.

I did:

- reset CMOS settings
- cleaned all fans and cleared of dust
- reseated graphics card, attached 6 pin GPU power connector to another 6-pin power slot in the PSU
- uninstalled driver 344.48 through DDU, and downgraded to 344.11 (sans the GeForce Experience, NVIDIA 3d vision, etc)

Now, it's different: running BF4/DOTA 2/Alan Wake will suddenly show temperature spikes of up to 100C. The game almost always slows down to a crawl when this happens. This is now happening EVERY time, though the game still continues to chug along, and in some cases, monitor shuts off again.

Is this a bad GPU? Is there anything else I can try? I don't have a spare GPU or spare anything to test unfortunately, except for a spare PSU. I'd appreciate your thoughts.

thanks.

 
Solution
No, if this is the first time you removed the Gpu heatsink

Try with an older driver,

If still no improvement, you might need a new one

ghost616

Reputable
Nov 5, 2014
16
0
4,510


The GPU is factory overclocked. I have tried down clocking but I'm not sure how you make it permanent.
 

ghost616

Reputable
Nov 5, 2014
16
0
4,510


I don't know how to improve the case airflow any further. All other system temps are OK (CPU, mobo chipset, HDDs). The video card is almost 2 years old, and I've never had issues until now.

I'll try other drivers; which NVIDIA ones would you recommend?
 

ghost616

Reputable
Nov 5, 2014
16
0
4,510


Thanks. I've done it before on an older graphics card, so I'll see if I can get it done again.
 

ghost616

Reputable
Nov 5, 2014
16
0
4,510
Update: reapplied thermal paste, same thing. Manages to go all the way up to 100C. I'll look for older graphics card drivers from NVIDIA and see if it helps.

I also noticed that the VRAM's didn't have thermal pads or coolers on them. Would that be a cause for concern?
 

ghost616

Reputable
Nov 5, 2014
16
0
4,510
I've tried 340.52. Same thing. I noticed that when it reaches 100C, it immediately throttles the GPU frequency to 542 mhz, hence the framerate drops.

I'm gonna try running this video card on another computer, maybe with a technician with lots of spare parts. I really want this to be just a video card issue, and nothing else.
 

ghost616

Reputable
Nov 5, 2014
16
0
4,510
Confirmed: I got a new video card (couldn't wait), and no overheating.

Makes me wonder what specifically was wrong with this card. Faulty VRMs, temp diodes? Wish I could send this over to someone to investigate.